An Empirical Investigation Of The Predictive Power Of Accrual And Cash Flow Data In Forecasting Operating Cash Flow
A primary objective of accounting information is to help financial statement users forecast the amounts, timing, and uncertainty of prospective cash inflows to the enterprise . While the Financial Accounting Standard Board (FASB) has expressed the belief that accrual accounting information is more useful than cash flow information in assessing the future cash flow prospects of an enterprise [8, 9], only limited empirical evidence is available on the ability of accrual and cash flow data to predict future cash flows.
In this paper, we address the issue of the relative information content (1) of facrual accounting and cash flow data with respect to the accuracy of cash flow forecasts. In addition, the possibly more important issue of whether a combination of accrual and cash flow data results in better predictions than utilizing only one data set is investigated. In so doing, the predictive information content of various components of accrual data (earnings, working capital, and sales) is examined.
While security price research has traditionally been the most popular means of evaluating the information content of accounting data, Beaver suggests that one means of testing the FASB's contention of earnings superiority "is to look at the forecasting ability of earnings and current cash flows with respect to future cash flows" [3, p. 129]. Security price research focuses on the role accounting data play in setting security prices. There is, however, a broader class of decisions that are affected by cash flow expectations. The informational demands of investors for a variety of decision contexts include projected cash flows . Lending institutions and other creditors are interested in the cash flow prospects of borrowers in order to assess the probability of collection. Investors require information about the future cash flows of investees to assess dividend prospects. Indeed, "a firm's abnormal rate of return [on its securities] . . . is expressed as a positive function of the unexpected cash flows of the period and the change in that period in the expected cash flows for future periods" [20, p. 65]. Assessing the relative predictive information content of accounting data allows for evaluation of these data without specifying a particular decision context [1, 4]. An investigation of the ability of accounting data to predict future cash flow encompasses a broader view of information content than does security price research.
To date, only one study reports evidence on the relationships between various accrual data and cash flows . The study concludes that random walk cash flow prediction models generally perform as well as models utilizing accrual data. Furthermore, it is stated that the results "do not support the FASB's assertions that earnings provide better forecasts of future cash flows than do cash flow measures" [5, p. 724].
The relationship of cash flow data with security prics has been more widely investigated [6, 16, 21, 22]. The focus of these studies is on the information content of accrual and cash flow data within the context of how these data are reflected in security prices. Rayburn  studied the relative association of accrual accounting components (current, long-term, and aggregate) and cash flow data with security returns. She found that operating cash flow, current accrual, and aggregate accrual data possess significant explanatory power, but that long-term accural data do not.
Wilson [21, 22] reported on the relative information content of accrual and funds flow data in two different studies. In one study, Wilson  investigated the information content of the current and noncurrent component of earnings by comparing the reaction of security prices surrounding the earnings announcement date to the reaction of security prices surrounding the date on which the annual report arrived at the SEC. …